I. INTRODUCTION
3.4. Research Instruments
3.4.1. Data Collection
3.4.1.3. Questionnaire
The researcher used structured-type questionnaire to gather the data as the respondents had to choose from the options specified by the researcher, or usually known as closed-end question. It is a question for which participants choose from a limited number of alternatives (Jackson, 2009). As previously mentioned, the researcher constructed the questionnaire based on four out of nine indicators of product differentiation. The total number of questions before doing validity and reliability test was 25 questions. The researcher spread the questionnaire by coming to the store of local producers. The questionnaire was made in Bahasa Indonesia. Moreover, rating scale used in this research was Likert rating scale. Likert rating scale is a special type of the more general class of summated rating scales constructed from multiple ordered-category rating items (Lavrakas, 2008).
Furthermore, Likert scale is a measurement scale with five response categories which ranging from “strongly disagree” to
“strongly agree”, in this rating scale, the respondents need to
President University | 2012 International Business Thesis 30 measure or indicate a level of agreement or disagreement for each statements related to the object. The value of Likert Scale was shown as below:
Value of Likert Scale Meaning
1 Strongly Disagree
2 Disagree
3 Neither Agree Nor Disagree
4 Agree
5 Strongly Agree
Table 3.4.1.3.: Likert Scale Source: Lavrakas (2008)
After all the data has been collected, the next step was to transform ordinal data into interval data in order to represent and determine one object has more value or characteristic comparing to other objects. To transform the data, the researcher used Successive Interval Method. Tool used by the researcher was Successive Interval Method (MSI) that was downloaded in (Juliandi). The tool assisted the researcher to instantly transform ordinal data into interval data.
3.4.2. Data Analysis
There are two tools used to analyze the research:
SPSS (Statistical Package for Social Science) version 16.0
Microsoft Excel 2007
The researcher combined both SPSS version 16.0 and Microsoft Excel 2007 to perform validity and reliability test, and also to process statistical data for the questionnaire.
President University | 2012 International Business Thesis 31
3.5. Sampling Design
3.5.1. Size of the population
The population used in this research was local sandal producers in Bali.
The researcher tried to find list of local sandal producers in Bali in several government institutions (e.g. Department of Trade and Industry).
Unfortunately, he could not find any data related to the research that concluded the size of the population was unknown. To decide numbers of sample used in the research, the researcher relied on theory developed by Roscoe Burwell Herrington (1975). He states that there are four rules to decide the number of sample:
1) The sample size more than 30 and less than 500 are appropriate for a lot of research;
2) If sample are divided into several categories (men/women, senior/junior, etc.) the sample size minimum 30 for each category;
3) In Multivariate research (include multiple regression analysis), the sample size will be better if 10 times or more much bigger than the variables on the research; and,
4) For simple experimental research with tight experiment control (match pairs, etc.) the successful research may use small sample size between 10 until 20.
As the research used factor analysis which was part of multivariate research, the researcher decided to follow rule number 3. The researcher multiplied numbers of variables used which were four by 10, so the minimum sample required was 40 respondents. However, factor analysis technique can only be implemented if sample used in a certain research is at least 50 respondents (Santoso, 2010). So, the researcher used 50 respondents for this research.
President University | 2012 International Business Thesis 32 3.5.2. Sampling Technique
Sampling technique used in this research was Snowball sampling technique. It is categorized as non-probability sampling technique.
Moreover, this technique is usually done if the researcher has inadequate information toward the population (Prasetyo & Jannah, 2010). Also, according to Lavrakas, (2008), the general objective of this technique is mainly to identify members of the rare population. It involves identifying one or more members of a rare population and asking them to name other members of the same population. The process continues until an adequate sample size has been obtained or until no new names are elicited from the process. In this research, the researcher used such technique to find other members of local sandal producers as the researcher was lack of information on the population. The technique was done until 50 respondents have been obtained on behalf of the research.
3.6. Statistical Treatment
3.6.1. Pilot TestingPilot test is also known as pre-test. Pilot tests are “dress rehearsals” of full survey operations that are implemented to determine whether problem exist that need to be addressed prior to putting the production survey in the field (Lavrakas, 2008). The objective of pilot test is to identify potential problems and address them prior to the production survey to reduce the amount of non-sampling measurement error produced by the survey. In this research, the researcher used two types of pilot tests, which are validity testing and reliability testing.
President University | 2012 International Business Thesis 33 3.6.1.1. Validity Testing
Validity test is test done in purpose to understand how accurate a certain instrument in measuring something (Priyanto, 2010). Validity is the extent to which a test measures what we actually wish to measure (Cooper & Schindler, 2006). The researcher conducted validity test by using SPSS version 16.0 program. The type of validity test used by the researcher was Corrected Item-Total Correlation Method. The validity of the questionnaire was determined by correlating between score of each item with the total score and conduct correction towards overestimated correlated coefficient value (Priyanto, 2010). The researcher used 20 respondents (n=20) as the sample of validity test, the calculation was n – 2 = 18 with significance level of 5%, so the value of r was 0.4438. Briefly stated, there are two rules in this method:
1) If the result r computation > rTable, the statement is valid; and, 2) If the result r computation < rTable, the statement is invalid.
Below table presented the result of validity test done by the researcher:
Table 3.6.1.1.1.: Result of Validity Test
Source: SPSS Version 16.0 and Primary data, Constructed by the Researcher
Variable Questions
Corrected Item-Total Correlation
rTable Remark
V1 Local sandals that I produce can be used for
less then one year. -.688 .4438 Invalid
V2 Local sandals that I produce can be used for
one to three years. .739 .4438 Valid
V3 Local sandals that I produce can be used for
more than three years. .633 .4438 Valid
V4 Local sandals that I produce are easy to clean. .637 .4438 Valid V5 Local sandals that I produce possess a good -.189 .4438 Invalid
President University | 2012 International Business Thesis 34 flexibility.
V6 Local sandals that I produce are water
resistant. .635 .4438 Valid
V7 Local sandals that I produce can withstand in
any field. .315 .4438 Invalid
V8 Local sandals that I produce are made by
sophisticated technology. .595 .4438 Valid
V9 Local sandals that I produce can be easily
repaired by the consumers. .022 .4438 Invalid
V10 I provide after-sale service. .631 .4438 Valid
V11 Local sandals that I produce need low cost to
repair. .621 .4438 Valid
V12 Materials used to produce the sandals are easy
to find. -.299 .4438 Invalid
V13 Local sandals that I produce do not need a long
time to be repaired. .592 .4438 Valid
V14 Local sandals that I produce are eye-catching. -.021 .4438 Invalid V15 I produce local sandals that are difficult to
imitate. .554 .4438 Valid
V16 I produce local sandals with a very fancy
packaging. .583 .4438 Valid
V17 Local sandals that I produce are very
comfortable. .055 .4438 Invalid
V18 Local sandals that I produce are very
prestigious. .735 .4438 Valid
V19 Local sandals that I produce are easy to wear. -.212 .4438 Invalid V20 Local sandals that I produce are customizable. .561 .4438 Valid V21 Some models of the sandals are produced in
limited edition. .596 .4438 Valid
V22 I always produce sandals that are up-to-date. .571 .4438 Valid V23 I produce sandals that do not make blisters on
feet. .656 .4438 Valid
V24 Local sandals that I produce are very colorful. .634 .4438 Valid V25 Local sandals that I produce reflect the culture
and characteristics of Bali. .650 .4438 Valid
President University | 2012 International Business Thesis 35 Based on the above table, there were 17 valid variables out of 25 variables. 8 variables were considered invalid as the result of r computation (Corrected Item-Total Correlation) did not exceed the rTable value which was 0.4438. So, the researcher eliminated invalid variables and used only 17 variables for the questions spread to respondents. Below table showed valid questions used by the researcher:
Table 3.6.1.1.2.: Result of Valid Variables
Source: SPSS Version 16.0 and Primary data, Constructed by the Researcher
Variable Questions
Corrected Item-Total Correlation
rTable Remark
V2 Local sandals that I produce can be used for
one to three years. .739 .4438 Valid
V3 Local sandals that I produce can be used for
more than three years. .633 .4438 Valid
V4 Local sandals that I produce are easy to clean .637 .4438 Valid V6 Local sandals that I produce are water
resistant. .635 .4438 Valid
V8 Local sandals that I produce are made by
sophisticated technology. .595 .4438 Valid
V10 I provide after-sale service. .631 .4438 Valid
V11 Local sandals that I produce need low cost to
repair. .621 .4438 Valid
V13 Local sandals that I produce do not need a long
time to be repaired. .592 .4438 Valid
V15 I produce local sandals that are difficult to
imitate. .554 .4438 Valid
V16 I produce local sandals with a very fancy
packaging. .583 .4438 Valid
V18 Local sandals that I produce are very
prestigious. .735 .4438 Valid
V20 Local sandals that I produce are customizable. .561 .4438 Valid
President University | 2012 International Business Thesis 36 V21 Some models of the sandals are produced in
limited edition. .596 .4438 Valid
V22 I always produce sandals that are up-to-date. .571 .4438 Valid V23 I produce sandals that do not make blisters on
feet. .656 .4438 Valid
V24 Local sandals that I produce are very colorful. .634 .4438 Valid V25 Local sandals that I produce reflect the culture
and characteristics of Bali. .650 .4438 Valid
3.6.1.2. Reliability Testing
Reliability test is done in order to test the consistency or stability of the instrument, whether the result will be consistent when the researcher repeatedly does the same measurement (Priyanto, 2010). A reliability coefficient is determined by assessing the degree of relationship between scores on the same test administered on two different occasions (Field, 2005). In this research, the researcher used Cronbach’s Alpha method. According to Sekaran, as cited in Priyanto (2010), Cronbach’s Alpha value less than 0.6 considered to be poor, while 0.7 is acceptable, and 0.8 or above considered good. Reliability test was done by inputting all valid variables to SPSS 16.0 program.
Table below showed the result of Reliability Test done by the researcher:
Cronbach’s Alpha N of Items
.915 17
Table 3.6.1.2.: Reliability Statistics Source: SPSS Version 16.0 and Primary data
President University | 2012 International Business Thesis 37 In the validity process, the researcher found out that there were 17 valid variables, so the researcher inputted 17 questions to seek out the value of Cronbach’s Alpha. From the above table, we could see that the value of Cronbach’s Alpha was more than 0.8 which was 0.915 that concluded the questionnaire was reliable as a research instrument.
3.6.2. Factor Analysis
A very concise definition of Factor analysis was stated by DeCoster (1998) that it is a collection of methods used to examine how underlying constructs influence the responses on a number of measured variables.
There are two types of variables in factor analysis:
1) Latent Variable
A latent variable is one which we do not observe directly, but we can infer its existence by the properties of observed variables (Jones). Latent variables or construct variables in this research were durability, reparability, style, and design.
2) Manifest Variable
According to DeCoster (1998), manifest variable is a variable that can be directly measured or observed. Manifest variables are used in latent variable statistical models, which test the relationships between a set of manifest variables and a set of latent variables.
Manifest variables are considered either continuous or categorical (a countable range). In this research, the manifest variables were details of each researched variables, for example eye-catching, difficult to imitate, aesthetics, packaging, and comfort as manifest variables of style.
President University | 2012 International Business Thesis 38 Latent variables, which are abstract concepts, have to be operationalized into manifest variables that can be measured into actual research.
Moreover, he explained that factor analysis can be categorized into two types: exploratory and confirmatory: Exploratory Factor Analysis (EFA) attempts to discover the nature of the constructs influencing a set of responses, while Confirmatory Factor Analysis (CFA) tests whether a specified set of constructs in influencing responses in a predicted way.
Moreover, factor analyses are performed by examining pattern of correlations (or covariance) between the observed measures. In this research, the researcher used Exploratory Factor Analysis (EFA) as the objective of the research is to determine number of common factors influencing a set of measures. Furthermore, adopting the seven basic steps to performing an EFA by DeCoster (1998), the steps done by the researcher are as follows:
1) Collect measurements: Measure variables on the same (or matched) experimental units. In this step, the researcher collected the data and summarized it in Microsoft Excel. Additionally, as the SPSS version 16.0 program obliged the researcher to input interval data, he had to transform ordinal data into interval data.
Tool used by the researcher was Successive Interval methods (Juliandi). Successive Interval Method helped the researcher summarizing the questionnaire based on each answer of each respondent.
2) Obtain the correlation matrix: Obtain the correlation (or covariance) between each of the variables. There were two steps to obtain the correlation matrix. The first step was The Bartlett’s Test of Spherecity which is used to test the null hypothesis that the original correlation matrix is an identity matrix. In this research,
President University | 2012 International Business Thesis 39 Bartlett’s Test is highly significant (p<0.001), and therefore the factor analysis is appropriate (Field, 2005).
The second step was Kaiser-Meyer-Olkin (KMO), it is a test to assess the appropriateness of using factor analysis on data. Kaiser, as stated in Field (2005) recommended retaining all factors with Eigen values greater than 1. The criterion is based on the idea that the Eigen values represent the amount of variation explained by a factor and that the Eigen value of 1 represents a substantial amount of variation.
3) Extract initial set of factors: Submit the correlations or covariance into a computer program to extract factors. A factor loading is the correlation between a variable and a factor that has been extracted from the data. Factor loading represents the correlation between original variables and its factors (Hair, Black, Babin, & Anderson, 2006). The researcher can employ the concept of statistical power to specify factor loadings considered significant for different sample size.
Loading Factors Minimum Sample Taken
.30 350
.35 250
.40 200
.45 150
.50 120
.55 100
.60 85
.65 70
.70 60
.75 50
Table 3.6.2.: Significant Factor Loading based on Sample Size Source: Hair, Black, Babin, and Anderson (2006)
President University | 2012 International Business Thesis 40 From the above table, the researcher could determine that the minimal factor loading for this research was 0.75, as the sample size used by the researcher was 50 respondents.
4) Rotate factors to a final solution: Find a factor solution that is equal to that obtained in the initial extraction but which has the simplest interpretation. Rotated factor loading is the factor loading for the varimax orthogonal rotation represent both how the variables are weighted for each factor but also the correlation between variables and the factor. A varimax rotation attempts to maximize the squared loading of the columns.
5) Interpret factor structures: Define a factor by considering the possible theoretical constructs that could be responsible for the observed pattern of positive and negative loadings. It is the step of analyzing the factors from the result of the rotation process.
3.7. Limitations
There were several limitations that the researcher faced during the finalization of the research:
1) The research scope was in the province of origin of the researcher, which was in Bali. Currently the researcher stays in Cikarang, so the researcher needed more time to collect all the data, both interview and questionnaire as respondents who became the primary data of the research were all originated from Bali. Also, several respondents were quite difficult to reach as they did not have enough complete addresses.
2) Some of secondary data needed by the researcher was not provided in governance institutions (e.g. lists of registered sandal producers in Bali), so the researcher had to find another source as the secondary data of the research.
President University | 2012 International Business Thesis 41 3) In this research, the researcher only analyzed the dominant factors of
product differentiation strategy that increase export potential chosen by the respondents, disregarding the personal information and background of the respondents (e.g. age, position, education, etc.)
4) The researcher also found that similar researches on specific dimensions of product differentiation were difficult to find. Though a lot of researches were implementing differentiation strategy, the researcher could only find a few researches that focus on specific dimensions of product differentiation.
5) Lastly, time limitation was also faced by the researcher that the research period conducted was only three months. Due to distance matter, the researcher needed more time to conduct the interview. Yet, the researcher could submit the research on-time with his best effort.
President University | 2012 International Business Thesis 42
CHAPTER IV ANALYSIS OF DATA
AND INTERPRETATION OF RESULT
4.1. Data Collection
As previously mentioned, after validity and reliability test have been done, the researcher found that there are 17 valid questions to be spread to the respondents. The researcher collected 50 questionnaires as the primary data of the research. Moreover, the researcher collected the questionnaires by directly coming to the office of local sandal producers, and some of them were reached by phone due to their far office location. Furthermore, the researcher could only take 50 respondents due to limitation of time and distance. Thus, 50 questionnaires are considered acceptable as Herrington (1975) stated in his research that in multivariate research, the sample size preferably several times (10 times or more) greater than the number of variables in the study. As variables of product differentiation used in this research were four (durability, reparability, style, and design), the minimum respondents taken for the research should be at least 40 respondents, and the researcher has exceeded the minimum requirement.
4.2. Research Variable
In this section, the researcher presents the table of construct variables used and the manifest variables to create the questions. The variable used in the questionnaire has already passed the validity and reliability testing. Table below shows the summary of valid questions.
President University | 2012 International Business Thesis 43 Table 4.2.: Table of Research Variables
Source: Primary data and Literature Study
4.3. General Information of the Respondents
In this section, the researcher presents the general information gathered from the respondents. The information was gathered from the questionnaire, by asking several questions related to sandal industries. The questions include office location, length of establishment, frequencies of design update, and product specialty of the respondents. Figures below explain information in details.
PRODUCT DIFFERENTIATION STRATEGY
Durability
V1 Local sandals that I produce can be used for one to three years.
V2 Local sandals that I produce can be used for more than three years.
V3 Local sandals that I produce are easy to clean.
V4 Local sandals that I produce are water resistant.
V5 Local sandals that I produce are made by sophisticated technology.
Reparability
V6 I provide after-sale service.
V7 Local sandals that I produce need low cost to repair.
V8 Local sandals that I produce do not need a long time to be repaired.
Style
V9 I produce local sandals that are difficult to imitate.
V10 I produce local sandals with a very fancy packaging.
V11 Local sandals that I produce are very prestigious.
Design
V12 Local sandals that I produce are customizable.
V13 Some models of the sandals are produced in limited edition.
V14 I always produce sandals that are up-to-date.
V15 I produce sandals that do not make blisters on feet.
V16 Local sandals that I produce are very colorful.
V17 Local sandals that I produce reflect the culture and characteristics of Bali.